New experimental characterisation methods for solid biomass fuels to be used in combined heat and power generation

NEW EXPERIMENTAL CHARACTERISATION METHODS FOR SOLID BIOMASS FUELS TO BE USED IN COMBINED HEAT AND POWER GENERATION

Gunnar Eriksson, Energy Engineering Div., Luleå University of Technology, S-971 87 Luleå, Sweden

Daniel Nordgren and Magnus Berg, Vattenfall Research & Development AB,

S-814 26 Älvkarleby, Sweden

ABSTRACT: The replacement of fossil fuels will lead to an increasing demand for unconventional biofuels. Fuel characterisation to predict combustion properties and facilitate the choice of combustion applications is important to avoid costly and time-consuming mistakes. Traditional methods are developed mainly for coal. Therefore procedures adapted specifically for solid biomass fuels are needed.

This work is a survey on approaches for combustion characterisation of biomass developed during the last ten years.

Innovative characterisation methods of interest concern:

1) Fuel handling behaviour: grindablility, erosion and abrasion properties.

2) Combustion characterisation: devolatilisation properties (important for ignition and flame stability), char burnout time.

3) Slagging and fouling properties of ash: ash particle formation, ash particle size distribution, ash composition, melting and gasification temperatures, slagging of bottom ash, reducing the risk by mixing with other fuels or using fuel additives and choice of suitable combustion applications for specific fuels.

The main conclusions are:

1) a method to measure grindability which takes electric power consumption into account is needed as the Hardgrove Grindability Index used for coal grinding is pointless for biofuels,

2) there is a need to develop convenient low-cost methods to measure slagging and fouling tendencies, devolatilisation kinetcs and char burnout for high heating rates found in fluidised beds and powder burners.

Keywords: agglomeration, ash, biofuels standardisation, biomass characteristics, biomass conversion, biomass drying, biomass/coal cofiring, boiler ash fouling, boiler ash, co-combustion, fly ash, fouling, slagging, solid biofuels

1 INTRODUCTION

More and more attention is directed towards increasing the share of renewable energy and biomass fuels are an important part of this in the short and medium run. Limits on land available to agriculture and forestry mean that productivity may have to be increased by switching to new energy crops or using new types of by-products. At the same time landfills are discouraged through taxes or even prohibited, creating a need to dispose of wastes and sludges elsewhere. Consequently many new types of biomass fuels may have to be used, in many cases unfamiliar to the energy plant operators. This creates a need to characterise combustion properties.

1.1 Relevant fuel properties

It can be assumed that an energy company faced with the decision whether to use a new fuel, proceeds in the following way (the process can be interupted at any stage if it turns out that the fuel is not suitable):

1. Economic assessment;

2. Visual assessment and use of background information about the fuel;

3. Chemical analysis (standard fuel analysis);

4. Advanced bench-scale tests;

5. Full-scale tests;

6. Updated economic assessment;

7. Long-term full-scale tests;

8. Updated economic assessment;

Even if the decision is not to use the fuel, it is reasonable to document the information gained, as there may be reason to reconsider the decision in the future.

Chemical analysis and the advanced bench-scale tests are commonly used to give information on some of the following:

1. storage and handling properties;

2. size distribution and grindability;

3. feedability;

4. combustion behaviour;

5. risk of ash related problems (slagging, fouling, corrosion, bed agglomeration (in the case of fluidised bed combustion);

6. emissions to be expected.

Rather then being inherent fuel properties, these issues are determined by a complex interaction between the fuel and the combustion equipment. As far as possible this study will be focused on the characterisation of fuel specific properties necessary to choose whether a particular fuel is suitable for a particular type of combustion equipment.

Properties like main elements, sulphur and chlorine content, main ash forming elements, trace elements, ash melting temperature during standard test conditions are often routinely measured. A short description of some of the conventional characterisation methods can be found at the webpages of national standardisation organisations like SIS [1] and the European standardisation organisation CEN [2]. An overview of methods can be found in the Fuel Handbook [3]

1.2. Objectives

The objective of the study was to summarise new experimental characterisation methods for biomass fuels to be used in coal-fired heat and power plants. Mainly laboratory- and bench-scale methods developed in the last ten years have been considered.

1.3 Method

The literature was searched for relevant information on advanced fuel characterisation methods. The EDTE database was used. The emphasis is on bench-scale experimental methods for fuel and ash characterisation.

The strategy was to search for general methods for characterisation of biomass fuels and solid recovered fuels rather than for characterisation of particular fuels.

The following limitations were used:

• Tar measurement techniques were not included since they were considered relevant for gasification rather than for combustion applications;

• Uses of ashes for purposes like construction, fertilisation etc were not included, the only concern for ashes was possible problems with combustion equipment;

• Only work published after the year 1997 were considered;

• Non-technical issues (e.g. economic and legislative issues) were not considered.

Papers fulfilling the search criteria but obviously irrelevant in this context were excluded (e.g. work on nuclear technology, geology, sewage treatment etc).

2 RECENT METHODS FOR FUEL CHARACTERI- SATION

2.1 Fuel handling, storage and feeding properties

2.1.1 Fuel sampling

A standardised sample preparation method for coal and biomass fuels was developed at the Energy Research Centre of the Netherlands (ECN) by drawing together various existing methods and applying new techniques [4].

2.1.2 Grindability

Grindability is mostly important for powder combustion.

In other combustion facilities fuel particles must be small enough to pass through lock hoppers, which means that it is necessary to crush the fuel.

Bergman et al at ECN have developed a novel method to determine the grindability of biomass. The net energy needed to break up the largest particles is measured [5].

2.1.3 Safety

Tests to determine explosive conditions for dust have been performed by Wilén and co-workers at VTT, Espoo, Finland [6]. A dust-air mixture is ignited inside a tank (20 litres or 1 m

) and pressure (1-25 bar) as a function of time is registered. Explosion parameters were measured at normal temperature and pressure, and at elevated

temperature and pressure. Lower Oxygen Concentration (LOC) decreases with increasing temperature, but increases slightly with increasing pressure.

Explosion suppression tests in a 1 m

vessel were also done by the authors. Monoammonium phosphate was blown into the vessel through two nozzles when an explosion was detected. Increasing temperature made suppression more demanding. Reducing the O

concentration to 17 percent made the suppression system significantly more efficient.

The risks of self-ignition of different fuels were quantified using TGA/DTA. The definition of thermal runaway used is a temperature increase of 50 K. The ignition temperature was defined as the temperature for which this happened. The following classification was used:

• Relatively inreactive dust (thermal runaway temperature above 400 ºC)

• Moderately reactive dust (thermal runaway temperature between 250 and 400 ºC)

• Most reactive dust (thermal runaway temperature below 250 ºC)

2.1.4 Fuel feeding

Bridging is a well-known problem when feeding biomass fuels, especially for straw and some other agricultural fuels. Paulrud et al designed a method to test bridging of powders using funnels with different opening sizes [8].

Another way to measure bridging of pulverised fuels was designed by Mattsson and Hofman [9]. The bottom gate of a commercial silo for storage of sand was used. A slot opening is gradually widened (using a pair of rolls) until the bridge of fuel particles is broken.

2.1.5 Fuel particle size and shape distribution

A method to measure particle size distribution and shape using optical microscopy has been developed at ECN [10]. An emulsion is created (to avoid density- induced particle segregation) between two standard microscope glass plates. Visible light is used to create a projection of the particles, which can be analysed for size and shape information. The more convenient laser diffraction methods assume spherical particles. Therefore the ECN method is most useful for larger biomass particles which are usually more non-spherical than smaller particles.

2.2 Combustion charateristics

A method to characterise biomass using a lab.scale

entrained flow reactor was developed by E. Biagini and

his collegues at Università di Pisa [11]. Thermogravi-

metric analyses, size measurements and SEM are used to

determine the conversion, reactivity and morphological

variations of solid residues for various operating

conditions. Models for fluid dynamics, energy balances

and heat and mass transfer were also developed. The

particles fed had diameters above 150 μm.

A method for determining the composition of an unknown waste mixture has been developed. The single components and the unknown mixture are characterised by a thermogravimetric analyser (TGA). It is assumed that the mixture TGA and (Differential Thermo- Gravimetric) DTA curves of the mixture are weighted sums of the curves of their respective components.

Synthetic four-component mixtures were used to test the method. The method works when the difference in decomposition temperature is in the order of tens of K [12]. Similarly, a method for using thermogravimetric analysis (TGA) for calculating compositions of biomass blends has been used on other fuel mixtures. Tests were performed on UK high volatile coal blended with palm kernel expeller, sawdust and olive cake. The devolatilisation profile were found to be additive with good accuracy [13, 14].

A method was developed to use TGA to characterise the de-volatilisation process of solid recovered fuels (SRF).

In combination with other well-established analytical procedures TGA is used to quantify the energy and elemental distribution between volatiles and char during the de-volatilisation process. The data can be used to compare SRF or its components with fuels like lignite and biomass [15].

The influence of minerals on devolatilisation kinetics was studied by Vamvuka [16]. The studied fuels were demineralised with acids. Raw and demineralised samples were analysed for ash content and their composition (elements and minerals), surface area and porosity were measured. Thermogravimetry from 25-80

o

C, at a heating rate of 10

C/min was used to study the reactivity of 250 µm particles. Changes in surface area and pore volume were measured. The result was that Ca, Mg, K and Si minerals generally acted as inert materials, inhibiting the pyrolysis and combustion rates of the samples, while lowering peak temperatures. According to the authors, mineral matter affected coal sample kinetics, while the influence on biomass sample kinetics was small.

One other study concerns the influence of alkali on devolatilisation kinetics. TGA measurements combined with GC/MS (Gas Chromatograph-Mass Spectrometer) measurements was used to study the effects of alkali on thermal degradation of biomass. A total of 19 Lolium and Festuca grasses (genetically mutated to give varying Klason lignin content). A strong catalytic effect, particularly from K, was observed both during pyrolysis and combustion. The char yield increases as the metal content (especially K and Na) decreases. Py-GCMS showed that peak intensities varied for untreated and treated samples; in particular the levoglucosan yield is higher and the hydroxyacetaldehyde yield is lower for treated (low metal content) samples. This supports previous work mechanisms by Liden et al. in which alkali metals promote an ionic route that favours ring- scission and hydroxyacetaldehyde formation [17].

The latent heat of pyrolysis was calculated by integration of differential scanning calorimetry (DSC) curves for biomass. A heating rate of 10 K/minute was used and the maximum temperature was 973 K. The results showed that 523 kJ, 459 kJ, 646 kJ and 385 kJ were required,

respectively, to increase the temperature of 1 kg of dried wheat straw, cotton stalk, pine and peanut from 303 K to 673 K [18].

For reactivity measurements, the Lab-scale Combustion Simulator (LCS, an entrained-flow reactor) at ECN, Netherlands, was re-designed to increase residence time from about 1 s to 2-3 s to allow char burnout (by reduction of the volumetric gas flow rate) [19].

To interpret results from a wired-mesh reactor, a mathematical model for the description of singe-particle pyrolysis (olive kernel particle) and prediction of product yields has been developed by A. Zabaniotou and coworkers. In the model kinetics (Koufopanos et al, two parallel first-order reactions) is coupled with a heat transfer model. The model has been validated against experiments in a laboratory wire mesh reactor, for a temperature range from 573 K to 873 K and a heating rate of 200 K/s [20].

A pyroprobe-GC/MS (Gas Chromatograph-Mass Spectrometer) was used to determine the composition of the thermal degradation products of lignin. Key marker compounds which are the derivatives of the three major lignin subunits (G, H, and S) were identified. For calibration, the marker compounds were matched with Klason lignin content using a partial least-square method.

The influence of alkali concentration on pyrolysis behaviour was studied. A total of 15 Lolium and Festuca grasses known to exhibit a range of Klason lignin contents were analysed [21].

A method for measurement of temperatures and gas composition in a burning fixed biofuel bed has been developed by Rönnbäck at SP, Sweden [22]. The influence of primary air flow, particle size and particle moisture content on the combustion can be studied. A batch-fired experimental rig using co-current feed of fuel and primary air was used. A temperature controlled probe to be tucked into the bed for gas analysis was constructed. During tests the fuel is ignited at the surface, the ignition front moving against the airflow and passing the probe on its way towards the grate. After the ignition front has reached the grate and the char combustion phase has been completed, CO, CO

, O

, CH

, THC and H

O are analysed. There are thermocouples at three different levels in the bed and primary air flow and weight change are continuously measured.

2.3 Ash-related problems and ash-related emissions 2.3.1 Extended fuel analysis

Scanning Electron Microscopy/Energy Dispersive X- ray analysis (SEM/EDS) can be used to discriminate between included and excluded minerals within the fuel matrix. Computer controlled SEM (CCSEM) has previously been used for coal characterisation. CCSEM has been adapted to biofuels at ECN [23]. The output is a normalised mass distribution of some 25 mineral types, divided over particle size intervals (2-4, 4-8, 8-16, 16-32, 32-64 and 64-128 µm). Ground biofuel is dispersed and embedded into resin. After hardening the resin block is cut, polished and carbon-coated for microscope analysis.

To avoid density segregation the resin holder was rotated

at low speed. CCSEM procedure at ECN has been updated using the new sampling procedures.

A simple method for identifying biomass fuels with problematic ash properties has been developed. The fuel is leeched in water and the conductivity of the solution is measured. A reasonably good correlation between conductivity and sintering temperature was found [24].

Practical tests have been done by Ena Energy, Sweden [25].

The extended fuel analysis is a fractionation method that consists of sequential leaching of a solid fuel with water, ammonium acetate and hydrochloric acid. The method was used in combination with thermodynamic calculations by Zevenhoven-Onderwater et al to predict fouling and slagging behaviour of ashes from coal, a peat, forest residue and Salix. Results from the fractionation showed clear differences in mineral distribution in the fuels [26].

2.3.2 Alkali release

Alkali is of importance for ash related problems like slagging and depositions, and bed agglomeration in fluidised beds. With non-homogeneous fuels, on-line methods for alkali measurements are important. Alkali is important in particle formation (see section 2.3.4).

The Pressurised Entrained Flow Reactor at VTT was used for studying alkali release from different coals. The detection method used was Plasma-Excited Alkali Resonance Line Spectroscopy (PEARLS) developed by Tampere University of Technology, where the alkali content not only in the gas phase but in fine particles is detected. Three kinds of coal were studied. The result was that the vapourisation of alkali depends stongly on particle temperature, and that high Cl content promotes alkali vapourisation [27].

Another method to determine the alkali concentration is by fluorescence induced with an Excimer laser (ELIF) and this has been showed by Glazer [28]. The alkali compound concentrations in the flue gas from coal co- combusted with four kinds of straw with different alkali content in a circulating fluidised bed) were measured.

Capture of KCl in a fixed-bed reactor was studied with an on-line alkali detector. The detector is based on surface ionisation and capable of detecting concentrations in the order of 1 ppb. Kaolin was added at a temperature of 850 ºC to capture alkali released from the fuel, and found to remove up to 99 percent of the alkali in the gas phase. When tests were made with wood at 650 ºC alkali release increased during pyrolysis because of impurities in the additive, alkali release decreased by about 50 percent during char combustion, the net effect being an approximate decrease in alkali release of about 20 percent [29, 30].

A new equipment for measuring alkali release from solid samples has been developed by Korsgren [31]. Alkali metals emitted from a sample are ionised at a hot platinum filament. The filament is heated to the temperature required to produce ionic vapour from alkali.

The ion current is proportional to the arrival rate of alkali

metal compounds. Temperatures up to 1 000 ºC and pressures up to 10 bar and either reducing or oxidising atmosphere are possible. Initial tests were done in a reducing atmosphere with reactors at various sizes from mg per batch to continuous feeding of kg/h [32]. Tests show that one fraction of the alkali is released below 500 ºC due to the breakdown of the organic structure.

Another fraction is released during char combustion. In biomass, a high chlorine content increses alkali release from the char [33]. The sensitivity to both Na and K was similar regardless of chemical environment. The response time was approximately 1 ms. Particles below 5 nm melt completely on the hot platinum surface. For larger particles the ionisation efficiency becomes lower and depends on the type of salt. Different filament configurations can be used to overcome this problem. A high-temperature aluminia-tube flow reactor was used by Dayton et al to study alkali metal release from coal- biomass mixtures under oxidising conditions using a MBMS [34]. Potassium chloride concentrations in the gas phase were lower than expected and the authors conclude that the reason was interaction with minerals in the coal. The combustion behavior, gaseous emissions, and alkali metals released during the combustion of several biomass/coal blends were investigated using a direct sampling, molecular beam mass spectrometer (MBMS) system in conjunction with a high-temperature alumina-tube flow reactor.

2.3.3 Ash melting properties

Simultaneous thermal analysis (STA) is used to characterise thermal changes as a function of temperature [34]. Thermo-gravimetric analysis (TGA) and differential scanning calorific analysis (DSCA, where the sample temperature is compared to the temperature of an inert reference material) are combined.

Frandsen et al have also used a method known as High Temperature Light Microscopy (HTLM, i.e. detecting phase changes of an unpressurised ash sample with a stereo microscope and personal-computer real-time image analysis). For the tests simple salt mixtures, geological standards and samples from straw and coal- straw pulverised fuel combustion were analysed. An N

atmosphere and a heating rate of 10 ºC/minute were used.

The methods are claimed to be more reliable than the ones previously available. An additional advantage is that they give information about the molten fraction as a function of the temperature rather than a single temperature value [35]. The method is also known as the Melt Area Fraction (MAF) method [37].

Ash fusibility has also been studied by Frandsen et al by measuring the viscosity of melted ashes [38]. Ashes are pre-treated: salt is washed off, residual carbon is burnt off and the ashes are pre-melted in a separate crucible (for volume reduction, controlled temperature history and homogenisation).

Work on standardisation of ash melting measurements is

reviewed by Hofbauer [39]. Two melting measurement

methods for biomass ashes were identified. One is based

on the existing standards used for coal ashes, the other is

novel method based on "Melt Area Fraction", developed

by dk-Teknik. The first method was adopted since it was already in use by several laboratories.

2.3.4 Particle formation

Fly ash particles from combustion cause ash deposition on material surfaces, which increases the temperature difference between the gas and water/steam side and may lead to corrosion. Combustion generated particles are also a potential health problem.

Formation and evolution of aerosols at high- temperatures have been investigated by Wiinikka et al [40]. An 8 kW updraft fired wood-pellets combustor was used. Particle samples were withdrawn form the centerline through 10 sampling ports by a rapid dilution sampling probe (temperatures from 200 to 1450 ºC). A low-pressure impactor was used to separate the particles according to size, and the chemical composition was studied using SEM/EDS and XRD. Results were compared to equilibrium calculations for aerosol formation. The particle size distribution had two peaks below 2.5 μm. The particle concentration as a function of position/residence time decreased at first due to burnout of carbon, thereafter it increased because of condensation of alkali sulphates, alkali chlorides and Zn species on existing particles, which agrees will with theory.

Particle formation has also been studied at Denmark University of Technology, Lyngby by Balzer Nielsen [41]. The work include:

1) Design of a lab-scale Laminar Flow Aerosol Condenser (LFAC)

2) Development of a mathematical model for formation and evolution of aerosols, implemented in FORTRAN, 3) Characterisation of aerosols from the Studstrup power plant, Århus, Denmark (3 types of coal and straw, submicron particles sampled in the economiser at 350 ºC using a novel ejector probe) using SMPS (Scanning Mobility Particle Sizer) and a Berner-type low pressure impactor, manual SEM study of morphology and chemical analysis and EDS to measure composition.

A method for estimating the characteristic contribution from various fuels to ambient PM 2.5 particles has been developed. It applies dilution sampling, cooling and humidification of stack gas to promote plume simulation, before particle analysis. The method has been used for several types of petroleum oils, bitumen emulsions, biofuels and pulverised coal blends, with different types of combustion equipment [42].

The Electrostatic Precipitator Test Facility at the Nottingham Fuel&Energy Centre, Univ. of Nottingham, has been built to simulate the conditions in a full-scale combustion plant [43]. The objective has been to improve understanding of the fundamental mechanisms involved in ESP flue gas cleaning, especially in the final stage when fly ash concentration is very low.

2.3.5 Slagging, fouling and corrosion

To compare different superheater materials during co-firing during 1000 h laboratory corrosion tests have been carried out by N.J. Simms [44]. Tests were done using controlled atmosphere furnaces at five superheater

alloy samples (1 Cr steel, 2.25 Cr steel, X20CrMoV121, AISI 347H and alloy 625) at temperatures typical for metal surfaces at superheaters and evaporators. The metal surfaces were coated with KCl, K

SO

and fly ash in different mixtures, to simulate deposits actually observed on superheaters. Mass loss was monitored.

Material performance was determined from dimensional metrology before and after exposure. SEM/EDX analyses of selected samples have been used to confirm whether changes in corrosion mechanism were associated with changes in damage levels (e.g. pitting to internal corrosion).

Submicron particles from small-scale equipment has been characterised by Johanssson et al [45] using ICP- MS (Inductively Coupled Plasma Mass Spectrometry), IC (Ion Chromatography) and TOF-SIMS (Time-of-flight Secondary Ion Mass Spectrometry), in addition to low- pressure impactor. Formation of inorganic particles was studied in a lab-scale reactor. A time-resolved signal of metal release from combustion of single pellets was obtained by ICP-OES (Inductively Coupled Plasma - Optical Emission Spectrometry). A photoelectric aerosol sensor was used as a soot indicator. The result was that most of the metal was released during char combustion.

A prediction method for deposit formation on heat exchanger surfaces in fluidised bed combustion is being developed. Experiments have been done at VTT in Jyväskylä; at Kvaerner Pulping in Tampere and at UPM Kymmene in Kaipola. A mixture of spruce bark, peat and forest residue was used. Deposit samples were analysed with SEM/EDS and a global equilibrium analysis was done for 500 - 900 ºC [46].

An air-cooled probe has been used to study deposits.

Computer-Controlled Scanning Electron Mictroscopy (CCSEM) has been used to detect changes in species composition for deposits, fly ashes and bottom ashes.

Silicates with K, Ca and Fe were found in the deposits.

Simultaneous Thermal Analysis (STA) was used to measure melting temperatures. No significant difference was found between fly ashes and bottom ashes. First melting according to STA started at temperature 150 ºC lower than the initial deformation temperature [47].

A pilot-scale (50 kW) Circulating Fluidised Bed (CFB) has been used by Vainikka to characterise fuels which are mixtures between coal, bark, peat and logging residues [48]. Feeding of limestone to reduce SOx emissions can lead to increased HCl production and possible corrosion problems. The "effective" S/Cl ratio, defined as the molar ratio in the gas phase, was found to be a relevant control parameter (while the total S/Cl molar ratio can be misleading. Comparisons with full- scale tests at Alholmens Kraft are planned.

A method for characterisation of fuels according to their

slagging properties in grate combustion has been

developed by Öhman et al [49]. A residential pellet

combustor is used to simulate conditions in a large-scale

grate combustor. Slag deposits are classified visually

according to the following four categories: 1) non-

sintered ash residue, 2) partially sintered ash, 3) totally

sintered ash (smaller blocks) and 4) totally sintered ash

(larger blocks). The composition of slag, non-sintered ash

and flyash (classified according to particles size) was studied using SEM-EDS and XRD, as described in 1.2. A database with about 50 fuels has since been compiled.

2.3.6 Agglomeration tendencies in fluidised beds

A method for quantification of bed agglomeration tendencies of different fuels had been developed and evaluated by Öhman and Nordin [50]. A bench-scale fluidised bed (5 kW) is used. Bed temperature is increased by an external heat to the primary air and to the bed section walls. In addition, temperature homogeneity is secured by switching from normal fuel feeding to a propane precombustor. The initial agglomeration temperature is determined by on- or off-line principal component analysis of the variations in measured bed temperatures (four values) and differential pressures (four). The agglomeration temperature of the fuel could be determined to 899 ºC (avg) with a reproducibility of {+-} 5 ºC. At TPS, Sweden a different measurement procedure is used. The fluidised bed is kept at constant temperature and the ash content is gradually increased until agglomeration occurs. Various temperatures have to be used for fuels with unknown properties [51].

A fluidised bed monitoring method is presented, based on pressure variations [52]. Ashes from co-combustion of recovered fuel (RF) and coal, peat, wood or wood-waste in a 15 kW fluidised bed were studied. Fly ashes were analysed using XRF. The sintering properties of the ashes were analysed with a test procedure developed at Åbo Akademi, Finland. Sintering was significant below 600 ºC and above 800 ºC for RF/wood and RF/bark, but not for RF/peat. Ash pellets were thermally treated with nitrogen to avoid residual carbon combustion. An increased level of alkali chlorides and sulphates made sintering increasingly likely [53] [58]

Larfeldt and Zindtl have tested a procedure for measuring bed agglomeration tendencies in a rotary furnace [54].

The heating of bed material from a fluidised bed in a rotary furnace is video recorded and the temperature is measured with an IR-sensor. The temperature when the molten sample is starting to stick to the sides of the crucible can thus be determined. For comparison the authors have made experiments in a bench-scale fluidised bed, and found that the temperature when 50 percent ash stuck to the sides in the rotary oven is approximately the temperature when agglomeration can be expected in a fluidised bed.

2.3.7 Function of SCR catalysts

Deactivation of catalysts for selective catalytic reduction (SCR) by aerosols has been studied by Frandsen [55]. Two bench-scale reactors for studying deactivation of SCR catalysts during biomass combustion have been constructed and tested. Full-scale SCR elements can be exposed to synthetic aerosols of well- defined composition. A reactor model has been developed and showed good agreement with measurements.

Kling and co-workers has developed a bench-scale reactor for testing the deactivation of SCR catalysis for different fuel mixtures. The temperature of the reactor is

controlled and normal operation temperatures are 250, 275, 350 and 400

C. A small sample of a catalyst is exposed to synthetic flue gas composition made of NO, NH

and O

. Conclusions from a test campaign with co- combustion of used wood and wood chips are that alkali poisoning is the main cause for deactivation and that dioxin concentrations can be reduced by 70 percent by a full-scale SCR catalyst. [56].

3 DISCUSSION

The literature search is limited to work published in 1998 or later were and relevant research may have been excluded because of the search criteria. Since the amount of work in the area is huge, it was necessary to limit the scope of this study.

Any characterisation method is a trade-off between accuracy, time and costs. While routine tests are well defined and performed in a structured and controlled manner, thermal and chemical conditions may be different from actual applications. Full-scale tests provide realism but may be costly, problematic and time consuming. The challenge is to design lab- and bench- scale methods which reproduce the main features of the full-scale energy plant.

The conditions in different types of combustion applications are considerably different. When grate combustion is used, fuel particles can be large. Thermal and chemical conditions are altogether different in each layer of the bed. Heating and drying take up considerable time. Fuel particles used in a powder burner typically have diameters below 1 mm. They are heated several orders of magnitude faster than particles on a grate. In the powder burner case, feeding and stable combustion may be a problem for some fuels. Grinding the feedstock to powder can also be problematic. During handling, dust explosions are a potential problem. In a fluidised bed, particles are of intermediate sizes, about 10 to 100 mm.

Thermal and chemical conditions are rather homogeneous for all particles regardless of their stage of combustion. Chemical interaction with the bed material is another complicating property. When considering the use of a fuel in a fluidised bed, some knowledge of the likelyhood for bed agglomeration temperature is important.

In these three cases, the distribution of the ash-forming elements between the different end products (fly ash, bottom ash, coarse particles, deposits at different locations in the boiler, gas phase.... ) is bound to be completely different, and so are the chemical and thermal conditions they go through. Therefore, fouling and slagging problems must be considered separately for each individual combination of fuel and combustion application.

Most characterisation tests can only reproduce some of these features. By choosing the conditions that are relevant for each particular case, data may still be useful in validated model for the properties to be predicted (e.g.

combustion, ash forming, deposition). .

The respective advantages and disadvantages of the

methods previously described will be considered in this

context for the three problem areas 1) Handling, grinding and feeding 2) Combustion properties, 3) Slagging and fouling.

Fuel handling, grinding and feeding

Concerning fuel storage and handling, grinding and feeding not so many results were found through the literature search.

The Hargrove Grindability Index which measures the mass fraction of particles which pass a sieve of 74 μm in pre-treated fuel, is not practically useful for biomass fuels. For most fuels, a high HGI can usually be achieved by increasing the grinding power used, an important parameter not taken into account by the method. A grindability index which does take power consumption into account is thus needed. One suggestion is the one already mentioned deviced at ECN [57], another has been proposed by Berg and his co-workers [58].

Combustion characteristics

There are a large number of recent methods available for characterising combustion properties. Heating rates have been shown to influence devolatilisation kinetics considerably, and this influence must always be kept in mind. When TGA analysis and related methods are used, the low heating rates make the results most relevant for grate combustion. With drop tube reactor methods heating rates similar to conditions in a powder burner can be achieved. With wired-mesh reactors high heating rates are also possible, although a drawback for powder fuels is the difficulty to discriminate between the intrinsic kinetics of the fuel and the heat and mass transfer properties, the latter depending on the experimental situation e.g. the size and shape of the entire fuel sample.

For modeling combustion of pulverised fuels, models combining fluid dynamics, energy balance and heat and mass transfer (e.g. the one developed by Biagini et al.

[59]) are useful, when combined with drop tube/entrained flow reactor measurements. For powders, data on particle size and shape distribution methods are important. The optical method from ECN should be useful for fuels with coarser particles but for powders there is a lack of simple and reliable methods.

There is a lack of convenient equipment which operators of boilers with powder burners can use on-site to get data on devolatilisation kinetics and char burnout rate with . TGA is widely used, no doubt because it is an accepted and relatively cheap method compared to drop-tube reactors/entrained-flow reactors. The obvious problem is the low heating rate (tens of K per minute), which can be an acceptable approximation for grate combustors but is several orders of magnitude lower than the conditions in a fluidised bed, and even more different from conditions in a powder burner. On the other hand, entrained-flow reactors require substantial investments and preparations for measurements are time-consuming. For most heat and power plant owners they are not available in-house and measurements must be done somewhere else in campaigns, planned well in advance. This probably limits their use. There is thus a need for some bench-scale

equipment of intermediate complexity and high heating rates and that preferably give data within the same day.

Ash-related problems and ash-related emissions

The methods for alkali release measurement seem to be too complex for routine use in bench-scale applications. There is certainly a possibility that they can be further developed and made more user-friendly, but currently they should be most useful for larger laboratories or for monitoring at full-scale units.

Grates and fluidised beds in bench-scale can be used to simulate fouling and bed agglomeration in full-scale facilities. The idea of using experimental grate combustor rigs like the one at SP should have potential for testing fuel-specific properties for grate combustion. On the other hand powder burner flames cannot be down-sized to bench-scale without affecting the properties studied, like flame stability. Measurements of relevant fuel- specific combustion properties in combination with modelling must therefore be used.

Concerning slagging and fouling problems, there is a need for bench-scale methods that are more similar to actual thermal and chemical combustion conditions,. A limitation of standard tests like the ASTM ash fusion test is that the conditions, while well defined are not very realistic. In one study, results from the ASTM standard ash fusion test and a compression strength based sintering test using laboratory ashes where compared with a continuously fed bed agglomeration (CFBA) test.

The ASTM standard ash fusion tests generally showed 50 - 500 ºC higher temperatures than the sintering test and the CFBA test. The sintering test generally showed 20 to 40 ºC lower sintering temperatures than the CFBA test [60].

Leaching and chemical fractionation could possibly be interesting to develop further as a biomass fuel characterisation tool. However it has not been systematically tested how well the results agree with thermal measurement methods.

The water leaching method, combined with conductivity measurements, could be used in combination with other methods as one early indication of possible alkali problems.

The Melt Area Fraction method could be useful as a complement to SEM/EDS and XRD measurements, provided that the ash material under study are produced during conditions which are relevant for the application studied.

Conclusions

In the literature search, very few results were found on grindability. Suggestions have been made for an index useful for biomass fuels instead of the Hardgrove Grindability Index, which is not practically useful for biofuels.

On combustion characterisation, most work used TGA,

which have slow heating rates suitable to simulate grate

combustion but not powder burners. To achieve heating

rates similar to conditions in powder burner, drop tube/entrained flow reactors or electric grid reactors are needed. Few such studies were found. One possible cause is that current methods are not sufficiently economical and convenient to be used routinely for new fuels.

There has been work where chemical methods instead of thermal methods were used to characterise fuel and ash properties. Possibly they could be more economical and convenient than thermal methods in some circumstances, provided that it can be established that they give the same relevant information.

Much work has been done on ash-related problems. Here, bench-scale methods exist for grate combustion and fluidised beds, but again there is a lack of methods to reproduce conditions in powder burners.

When considering co-combustion of biomass fuels in powder burners it can be concluded that there is a need for grindability measurement methods adapted to solid biomass fuels which (unlike the Hargrove Grindability Index) take electric power consumption (the property which differs between different biomass fuels) into account.

Leaching and chemical fractionation could be interesting to develop further as a biomass fuel characterisation tool.

A systematic comparison with thermal methods would be worthwhile.

The design of a convenient low-cost experimental device with high heating rates of the fuel to be tested for measuring slagging and fouling and/or devolatilisation kinetics and/or char burnout rate should be a subject for future studies.

ACKNOWLEDGEMENTS

This work was sponsored by Vattenfall AB, Group Function R&D

4 REFERENCES [1] www.sis.se

[2] www.cen.eu/cenorm/homepage.htm

[3] Strömberg, Birgitta (2006): Fuel handbook, VÄRMEFORSK--971, Project Värmeforsk-F4- 324

[4] Cieplik, M.K. (2003): Improved procedures for sample preparation and CCSEM analysis, 12.

international conference on coal science, Cairns, Qld. (Australia), 2-6 Nov

[5] Bergman, P.C.A. et al (2004): Torrefaction for entrained-flow gasification of biomass, Contributions of ECN Biomass to the 2nd world conference and technology exhibition on biomass for energy, industry and climate protection, Rome, 10-14 May 2004, ECN report number ECN-RX-- 04-029

[6] Wilén, C. and Rautalin, A.(1999): Safety aspects of fuel handling in IGCC and PFBC plants, 15th International Conference on Fluidized Bed Combustion, Savannah, GA (US), Research

Org:VTT, Jyväskylä Science Park, Jyväskylä (Finland)

[7] Sun, Jin (2007): Multiscale modeling of segregation in granular flows, PhD dissertion, Ames Laboratory, Ames, IA,

[8] Paulrud, Susanne (2004): Upgraded biofuels - effects of quality in processing, handling characteristics, combustion and ash melting, AgrD thesis, Swedish Agricultural University, Umeå, Sweden

[9] Mattsson, J. E. and Hofman, P.D. (2002): Method and apparatus for measuring the tendency of solid biofuels to bridge over openings, Biomass and Bioenergy 22, 179-185

[10] Korbee, R. et al (2004): Cost-effective screening of biomass materials for co-firing, In:

Contributions ECN Biomass to the 2nd World Conference and Technology Exhibition on Biomass for Energy, Industry and Climate Protection

[11] Biagini, E. et al (2005): Development and characterization of a lab-scale entrained flow reactor for testing biomass fuels, Fuel 84 (12/13), 1524-1534

[12] Heikkinen, J. and Spliethoff, H.(2002):

Characterisation of waste fuels with TGA, VTT Technical Research Centre of Finland, Espoo (Finland), Technical Report VTT-SYMP--222 [13] Lester, Edward et al (2007): A method for source

apportionment in biomass/coal blends using thermogravimetric analysis, Journal of Analytical and Applied Pyrolysis 80(1), 111-117

[14] Gong, Mei et al (2007): Determination of coal biomass blend proportions using thermal analysis, 2007 international conference on coal science and technology, Nottingham (United Kingdom), 28-31 Aug

[15] Hilber, T. et al (2007): A method to characterise the volatile release of solid recovered fuels (SRF), Fuel, 86(1/2), 303-308

[16] Vamvuka, D. et al (2006): The effect of mineral matter on the physical and chemical activation of low rank coal and biomass materials, Fuel 85 (12/13) 1763-1771

[17] Fahmi, R. and Bridgwater, A.V.(2007): Prediction of Klason lignin and lignin thermal degradation products by Py-GC/MS in a collection of Lolium and Festuca grasses, Journal of Analytical and Applied Pyrolysis 80 (1) 16-23

[18] He, Fang (2006): Investigation on caloric requirement of biomass pyrolysis using TG-DSC analyzer, Energy Conversion and Management 47 (15-16) 2461-2469

[19] See ref. [10]

[20] Zabaniotou, A.et al (2007):

[21] See ref. [17]

[22] Rönnbäck, Marie et al (2003): The combustion process in a bed of biofuels II, Swedish Energy Agency, Eskilstuna (Sweden) Technical Report STEM-P808-6

[23] See ref. [10]

[24] Berge, N. et al (2003): The TPS Branch Research Program for District Heating Utilities 1999-2000.

Summary of projects, Swedish Energy Agency,

Eskilstuna (Sweden), Technical Report STEM-

FBT--03-13; SVF--755 ;STEM-P7270-6

[25] Gräns, Hilde and Larfeldt, Jenny (2001): The influence from fuel properties on deposit growth in biomass fired boilers, Technical Report SVF--727 / -F9-804, Värmeforsk, Stockholm (Sweden) [26] Zevenhoven-Onderwater, M. et al (2000): The

prediction of behaviour of ashes from five different solid fuels in fluidised bed combustion, Fuel 79 (11) 1353-1361

[27] Aho, M. (1998): Effects of fuel properties, temperature, and pressure on fuel reactivity, formation and destruction of nitrogen oxides, and release of alkalis, LIEKKI 2 Research Programme, Combustion and Gasification Research Programme. Technical Review 1993-1998; Hupa, M.; Matinlinna, J.[eds.] Åbo Akademi, Combustion Chemistry Research Group Turku (Finland)

[28] Glazer, M.P. et al (2006): Co-combustion of coal with high alkali straw. Measuring of gaseous alkali metals and sulfur emissions monitoring

[29] Steenari, Britt-Marie and Lindqvist, Oliver (2002):

Chemical Processes Related to Combustion in Fluidised Bed, Swedish Energy Agency, Eskilstuna (Sweden), Technical Report STEM- FBT--02-23, Project STEM-P12859-1

[30] Tran, Khanh-Quang (2004): On the application of surface ionization detector for the study of alkali capture by kaolin in a fixed bed reactor, Fuel 83 (7/8) 807-812

[31] Korsgren, J.G. et al (1999): Characterization of alkali metal emission from fuels and samples collected from fluidized bed gasification, 15th International Conference on Fluidized Bed Combustion, Savannah, GA (US In: Proceedings of the 15th national conference on fluidized bed combustion, by Reuther, R.B.[ed.]

[32] Wiktorsson, L.-P.A. et al (1998): Test of on-line alkli detector based on surface ionisation technique, EUR--18393/III-EN. Joule II - Programme. Clean coal technology R& D. 2nd phase. Volume III. Novel approaches in advanced combustion (pressurized systems); Hein, K.R.G.;

Minchener, A.J.; Pruschek, R.; Roberts, P.A.[eds.]

[33] See ref. [27]

[34] Dayton, D.C. et al (1999): Effect of coal minerals on chlorine and alkali metals released during biomass/coal cofiring, Energy and Fuels 13 (6), 1203-1211 (Nov-Dec)

[35] Frandsen, F.J. (1998): Characterization of ashes from biofuels, Energiministeriets Forskninsudvalg for produktion og fordeling af el og varme.

Braendsler og forbraendingsteknik, Technical Report NEI-DK

[36] See ref [35]

[37] See for instance: BIONORM (2004): Pre- normative work on sampling and testing of solid biofuels for the development of quality assurance systems. Work package II, Task 2, Ash melting behaviour, Final report, Contract ENK6-CT-2001- 00556, European Commission 5th Framework programme ENERGIE (December)

[38] Frandsen, F.J. (2002): Clean and efficient application of biomass for production of powser and heat - Phase 3 in a long-term strategic research project Danmarks Tekniske Univ., Lyngby

(Denmark). Inst. for Kemiteknik Technical Report, CHEC-R--0302, Contract ENS-1323/99-0001 [39] Hofbauer, H. (2004): Ash melting behaviour -

status of the standardisation, International conference on standardisation of solid biofuels:

Status of the ongoing standardisation process and results of the supporting research activities (BioNorm), Leipzig (Germany), 6-7 Oct 2004;

Related Information: In: International conference - Standardisation of solid biofuels. Status of the ongoing standardisation process and results of the supporting research activities (BioNorm).

ProceedingsInstitut für Energetik und Umwelt GmbH, Leipzig (Germany)

[40] Wiinikka, Henrik et al (2006): High-temperature aerosol formation in wood pellets flames: Spatially resolved measurements, Combustion and Flame 147, 278-293

[41] Balzer Nielsen, Lars (1998): Combustion aerosols from potassium-containing fuels, PhD thesis NEI- DK--3392, Inst. for Kemiteknik;Danmarks Tekniske Univ., Lyngby (Denmark)

[42] Win Lee, S. (2005): Particulate characteristics data for the management of PM2.5 emissions from stationary combustion sources, 6 international symposium on gas cleaning at high temperatures, Osaka (Japan), 20-22 Oct 2005, In: Advanced gas cleaning technology, by Kanaoka, C.; Makino, H.;

Kamiya, H. (eds.).

[43] Barranco, R. et al (2003): Studies into fly ash collection issues using a small-scale electrostatic precipitator test facility, 12th international conference on coal science. Coal - contributing to sustainable world development

[44] Simms, N. J. (2006): Heat exchanger corrosion in biomass and coal co-fired power plants, 8. Liège conference: Advanced materials for power engineering 2006, Liège (Belgium), 18-20 Sep 2006; In: Materials for advanced power engineering 2006. Proceedings of the 8th Liège conference. Pt. 3, Schriften des Forschungszentrums Jülich. Reihe Energietechnik/Energy Technology v. 53 Pt. 3, by Lecomte-Beckers, J.; Carton, M.; Schubert, F.;

Ennis, P.J. (eds.)

[45] Johansson, Linda et al (2004): Formation and emission of PM{sub 10} in combustion of biofuels. Final report, Swedish Energy Agency, Eskilstuna (Sweden) Technical report

[46] Zevenhoven, M.et al (2001): The chemistry and melting behaviour of fly ash deposits in co- combustion of bark, peat and forest residue, Technical Report, Åbo Akademi, Turku (Finland).

Process Chemistry Group

[47] Hansen, L.A. et al (1999): Characterization of ashes and deposits from high-temperature coal- straw co-firing, Energy and Fuels 13 (4) (Jul-Aug) [48] Vainikka, P. et al (2004): Maximum biomass use and efficiency in large-scale cofiring - PUUT31, in VTT Symposiumn 231, VTT Technical Research Centre of Finland, Espoo (Finland)

[49] Öhman, M. et al (2004): Slagging tendencies of wood pellet ash during combustion in residential pellet burners, Biomass and Bioenergy 27, 585-59 [50] Öhman, M.and Nordin, A (1998): A new method

for quantification of fluidized bed agglomeration

tendencies: a sensitivity analysis, Energy and Fuels 12( 1), 90-94

[51] Sjöström, Krister (2004): Gasification research at Royal Inst. of Technology July 2000 - Jun 2002, Swedish Energy Agency, Technical Report, FBT-- 04-6/P12852-1, Eskilstuna (Sweden)

[52] Ommen, J. R. et al (1999): An early-warning- method for detecting bed agglomeration in fluidized bed combustors, Proceedings, 15th International Conference on Fluidized Bed Combustion, Savannah, GA (US), Delft Univ. of Technology (NL)

[53] Zevenhoven, R. et al (1998): Characterization of ashes from co-combustion of refuse-derived fuel with coal, wood and bark in a fluidized bed, Proceedings 15. annual international Pittsburgh coal conference, Pittsburgh, PA (United States), (14-18 Sep 1998)

[54] Larfeldt, Jenny and Zindtl, Frank (2004): Tests of bed agglomeration tendency using a rotating furnace, Värmeforsk, Stockholm, Sweden, Technical Report F-208

[55] See ref. [38]

[56] Kling, Åsa et al (2005): SCR during co- combustion of biofuel and recycled fuels, Värmeforsk, Stockholm (Sweden), Technical Report VÄRMEFORSK--932, Project Vaermeforsk-F4-220

[57] See ref. [5]

[58] Berg, M. et al (DRAFT): New laboratory methods for the practical assessment of the burning characterisics of coal in large furnaces.

[59] See reference [11]

[60] Skrifvars, B.-J. et al (1999): Predicting bed

agglomeration tendencies for biomass fuels fired in

FBC boilers: A comparison of three different

prediction methods, Energy & Fuels, 13, 359-363